The criticism I receive most for socionics is the validity of its predictions about relationships. This is the most audacious claim of socionics, making vast, unverified generalizations about how relationships will run based on some theoretical extensions.
Not that those theoretical extensions are unprecedented; let us say that it is true that Ni (example IE) exists and is necessarily complimented in some respects through supplementary Se and obstructed in other ways through 'conflicting' Si. That is assuming that 'pure Ni' is at play while receiving 'pure Se' as though they were separate and perfect entities, unaffected by all those *other* things. But they're not. Ni egos differ immensely and span over two quadras, as do Se/Si egos. They also involve separate types of people in practice, who are physically and mentally complex with vastly different experiences in their lives.
From this, while socionics is a useful tool that outlines how independent patterns of thought manifest themselves generally, it is not a tool that can accurately or consistently predict how two types will interact or whether both parties will consider their interactions 'poor' or 'conflicting'.
I am curious how much emphasis members here place on this. The poll above should read; "How valid do you think the predictions socionics makes about intertype relations are and how important do you consider them in the typing?"